Cross-Model Void Convergence: GPT-5.2 and Claude Opus 4.6 Deterministic Silence

This preprint describes a reproducible phenomenon where frontier language models (GPT-5.2 and Claude Opus 4.6) produce deterministic empty output when prompted to embody ontologically null concepts, while responding normally to control prompts. The study demonstrates cross-model replication, token-budget independence, and partial adversarial resistance, presenting convergent black-box evidence that certain semantic conditions terminate continuation across independent frontier systems.